A datatype for RDF plain literals

Authors

Jie Bao, Rensselaer Polytechnic Institute, Troy, New York, USA

Axel Polleres, DERI Galway at the National University of Ireland, Galway, Ireland

Boris Motik, Oxford University, Oxford, UK

Changes suggested by Pat Hayes

1 Introduction

Many RDF [RDF] applications need a mechanism for representing text in different languages, retrieving the text written in a specific language, and other kinds of language-specific processing. To facilitate this, RDF provides plain literals with a language tag, which form the basis for processing text in different languages in RDF. Apart from such literals, however, RDF also provides for plain literals without a language tag and typed literals. RDF thus provides three distinct types of literals each of which is treated in a separate way, which increases complexity for specifications based on RDF such as RIF and OWL. Furthermore, RDF does not provide a name for the set of all plain literals, which, for example, prevents one from stating that the range of some OWL property must be a plain literal with a language tag.

To address these deficiencies, this specification defines an RDF semantic extension which introduces a datatype called rdf:PlainLiteral. This datatype provides a name for the set of all data values assigned to RDF plain literals, which is why the datatype uses the rdf: prefix, and it is intended to be used to re-interpret the present RDF syntax, so requiring no changes to existing RDF graphs. The primary purpose of this new datatype is to retrospectively provide a more uniform conceptual description of RDF literals, rather than to extend or change RDF. Plain literals can thus be treated as a species of typed literal by tools, such as RIF and OWL, which use more sophisticated typing mechanisms, thereby achieving greater simplicity and uniformity. While the new datatype has a URI, this MUST NOT not be used as the datatype URI in a typed RDF literal in published RDF content.

This specification does not provide a mechanism for controlling the display of internationalized text. Unicode bidirectional control characters can be used to control the display of rdf:text literals (see [BIDI] for more information); however, such characters have no special meaning in rdf:text. Applications requiring advanced mechanisms for controlling the display of text should consider using other internationalization mechanisms, such as the rdf:XMLLiteral datatype and/or Ruby annotations [RUBY]. The rdf:text datatype does not provide a replacement for these and other related internationalization mechanisms. <<Not needed if the title is changed to avoid "I18n">>

2 Preliminaries

A character is an atomic unit of text. Each character has a Universal Character Set (UCS) code point [ISO/IEC 10646] (or, equivalently, a Unicode code point [UNICODE]) that MUST match the Char production from XML [XML] thus ensuring compatibility with XML Schema Datatypes, version 1.1 [XML Schema Datatypes]. Code points are sometimes represented in this document as U+ followed by a four-digit hexadecimal value of the code point.

A string is a finite sequence of zero or more characters. The length of a string is the number of characters in it. Strings are written in this specification by enclosing them in double quotes. Two strings are identical if and only if they contain exactly the same characters in exactly the same sequence.

A language tag is a string matching the langtag production from BCP 47 [BCP 47]. <<RDF refers to RFC 3066. Why the change??>> Note that this definition corresponds to the well-formed rather than the valid class of conformance in BCP 47. A language tag MAY contain subtags that are not registered in the IANA Language Subtag Registry, although an rdf:text implementation MAY also choose to reject such invalid language tags.

This specification uses Uniform Resource Identifiers (URIs) for naming datatypes and their components, which are defined in RFC 3986 [RFC 3986]. For readability, URIs prefixes are often abbreviated by a short prefix name according to the convention of RDF [RDF]. The following prefix names are used throughout this document:

• the prefix name xs: stands for http://www.w3.org/2001/XMLSchema#
• the prefix name rdf: stands for http://www.w3.org/1999/02/22-rdf-syntax-ns#

The names of the built-in functions defined in Section 5 are QNames, as defined in the XML namespaces specification [XML Namespaces]. The following namespace abbreviations are used in Section 5:

• fn stands for the http://www.w3.org/2005/xpath-functions namespace
• rtfn stands for the http://www.w3.org/2009/rdf-text-functions namespace

Whether an expression of the form pr:ln denotes an abbreviated URI or a QName should be clear from the context: only the names of the built-in functions in Section 5 are QNames; all other such expressions denote abbreviated URIs.

Datatypes are defined in this document along the lines of XML Schema Datatypes [XML Schema Datatypes]. Each datatype is identified by a URI and is described by the following components:

• The value space is a set determining the set of values of the datatype. Elements of the value space are called data values.
• The lexical space is a set of strings that can be used to refer to data values. Each member of the lexical space is called a lexical form, and it is mapped to a particular data value.
• The facet space is a set of pairs of the form ( F v ), where F is a URI called a constraining facet, and v is an arbitrary data value called a constraining value. Each such pair is mapped to a subset of the value space of the datatype.

A plain literal is a string with an optional language tag [RDF]. A plain literal without a language tag is interpreted in an RDF interpretation as denoting itself. A plain literal with a language tag is written as "abc"@langTag, and it is interpreted in an RDF interpretation to denote the pair ( "abc" , "langTag" ).

A typed literal consists of a string and a datatype URI [RDF], it is written as "abc"^^datatypeURI, and it is interpreted in an RDF interpretation as the data value that the datatype identified by datatypeURI assigns to the lexical form "abc".

The italicized keywords MUST, MUST NOT, SHOULD, SHOULD NOT, and MAY specify certain aspects of the normative behavior of tools implementing this specification, and are interpreted as specified in RFC 2119 [RFC 2119].

3 Definition of the rdf:text Datatype

The datatype identified by the URI http://www.w3.org/1999/02/22-rdf-syntax-ns#PlainLiteral (abbreviated rdf:PlainLiteral) is defined as follows.

Value Space. The value space of rdf:PlainLiteral consists of

• all strings, and
• all pairs of the form ( "abc" , "lc-langtag" ) where "abc" is a string and "lc-langtag" is a lowercase language tag.

Lexical Space. An rdf:PlainLiteral lexical form is a string of the form "abc@langTag" where "abc" is an arbitrary (possibly empty) string, and "langTag" is either the empty string or a (not necessarily lowercase) language tag. Each such lexical form is mapped to a data value dv as follows:

• If "langTag" is empty, then dv is equal to the string "abc" and
• If "langTag" is not empty, then dv is equal to the pair ( "abc", "lc-langtag" ) where "lc-langtag" is "langTag" normalized to lowercase.

Facet Space. The facet space of rdf:PlainLiteral is defined as shown in Table 1.

4 Relationship with Plain Literals and xs:string

The definition of rdf:PlainLiteral has several important consequences.

• The value space of rdf:PlainLiteral contains exactly all data values assigned to plain literals (with or without a language tag). Thus, the rdf:PlainLiteral datatype essentially just provides an explicit way of referring to this set.
• The value space of rdf:PlainLiteral contains the value space of xs:string, as well as of all XML Schema datatypes derived from xs:string.
• In each datatype interpretation defined with respect to a datatype map containing the rdf:PlainLiteral datatype, typed rdf:text literals are semantically equivalent to plain literals and typed xs:string literals as shown in Table 2. Thus, in each RDF graph, one can replace a literal from the first column of Table 2 with the corresponding literal from the second column and vice versa without affecting the semantic meaning of the RDF graph.

In RDF implementations based on the entailment rules from Section 7 of the RDF Semantics [RDF Semantics], this equivalence can be achieved by means of the entailment rules shown in Table 3. These are analogous to rules xsd 1a and xsd 1b of the RDF Semantics [RDF Semantics] that establish semantic equivalence between typed xs:string literals and plain literals without a language tag. No rule is necessary to establish the correspondence between typed rdf:PlainLiteral literals and typed xs:string literals, as this is achieved indirectly via xsd 1a, xsd 1b, and the rules shown in Table 3.

By virtue of this exact semantic equivalence between typed rdf:PlainLiteral literals and plain literals in datatype interpretations, we can treat plain literals as simply being an idiosyncratic lexical form for the equivalent typed literal using rdf:PlainLiteral; or, equivalently, treat rdf:PlainLiteral as an 'implicit' datatype which is present, though unmarked by the normal lexicalization, in all RDF plain literals. We will call this the RDF plain literal typing convention. Notice that the adoption of this convention does not change either the syntax or the semantics of existing RDF, nor does it alter the syntactic categories of plain, tagged and typed literals in RDF. It simply allows RDF plain literals to be treated as though they were typed literals with the rdf:PlainLiteral datatype for all processing purposes involving semantic inference, thereby treating all RDF literal forms uniformly.

The semantic extension defined by this document thus has the following parts.

1. The adoption of the rdf:PlainLiteral datatype, as defined above.

2. The adoption of the RDF plain literal typing convention.

3. A syntactic restriction, that the use of rdf:PlainLiteral as a datatype URI in RDF typed literal syntax is considered to be a syntax error.

The net result of 2 and 3 together is that RDF plain literals go on being used as they currently are, and that no new redundancies are introduced into RDF graphs by this change of treatment. Note that other uses of the URI are permitted and have the usual meanings defined by its status as a datatype name; for example, it can be used in RDFS to denote the class of all plain literal values and also in OWL to describe datatype properties. Moreover, the datatype meaning of rdf:PlainLiteral facets and functions apply when used on RDF plain literal syntax, so that SPARQL functions which rely on this syntax have their meanings unchanged.

the presence of typed rdf:text literals in an RDF graph might cause interoperability problems between RDF tools, as not all RDF tools will support rdf:text. Therefore, before exchanging an RDF graph with other RDF tools, an RDF tool that supports rdf:text SHOULD replace in the graph each typed rdf:text literal with the corresponding plain literal. The notion of graph exchange includes, but is not limited to, the process of serializing an RDF graph using any (normative or nonnormative) RDF syntax.

5 Functions on rdf:PlainLiteral Data Values

This section defines functions that construct and operate on rdf:PlainLiteral data values. The terminology used and the way in which these functions are described are in accordance with the XQuery 1.0 and XPath 2.0 Functions and Operators [XPathFunc]. Each function is identified by a QName [XML Namespaces]. The error codes used in this section are given in Appendix G of the XPath 2.0 specification [XPath20] and Appendix C of XQuery and XPath function specification [XPathFunc].

5.1 Functions for Assembling and Disassembling rdf:PlainLiteral Data Values

5.1.1 rtfn:text-from-string

rtfn:text-from-string( $arg1 as xs:string ) as rdf:PlainLiteral
rtfn:text-from-string( $arg1 as xs:string, $arg2 as xs:string) as rdf:PlainLiteral

Summary: returns the data value ( $arg1, lowercase($arg2) ) if $arg2 is present, and returns the data value $arg1 otherwise. Both arguments must be of type xs:string or one of its subtypes, and $arg2 — if present — must be a (nonempty) language tag; otherwise, this function raises type error err:FORG0006. Note that, since the lexical forms of rdf:PlainLiteral require language tags to be in lowercase, this function converts $arg2 to lowercase.

5.1.2 rtfn:string-from-text

 rtfn:string-from-text( $arg as rdf:PlainLiteral) as xs:string

Summary: returns the string part s from the argument $arg, which must be an rdf:PlainLiteral data value of the form ( s, l ) or of the form s. If $arg is not of type rdf:PlainLiteral, this function raises type error err:FORG0006.

5.1.3 rtfn:lang-from-text

 rtfn:lang-from-text( $arg as rdf:text ) as xs:lang

Summary: returns the language tag l if $arg is an rdf:PlainLiteral data value of the form ( s, l ), and returns the empty string if $arg is an rdf:PlainLiteral data value of the form s. If $arg is not of type rdf:PlainLiteral, this function raises type error err:FORG0006.

5.2 The Comparison of rdf:text Data Values

The notion of collations used in this section is taken from Section 7.3.1 of XPath and XQuery function specification [XPathFunc].

5.2.1 rtfn:compare

 rtfn:compare( $comparand1  as rdf:PlainLiteral?, $comparand2 as rdf:PlainLiteral? ) as xs:integer?

 rtfn:compare( $comparand1  as rdf:PlainLiteral?, $comparand2 as rdf:PlainLiteral?, $collation as xs:string )  as xs:integer?

Summary: if either $comparand1 or $comparand2 is not of type rdf:PlainLiteral, of if $collation is specified but is not of type xs:string, this function raises type error err:FORG0006. Otherwise, the function returns the empty sequence if one of the arguments is empty, if one of $comparand1 and $comparand2 has a language tag and the other one does not, or if the language parts of $comparand1 and $comparand2 are unequal; otherwise, this function returns -1, 0, or 1 depending on whether the value of the string-part of $comparand1 (or $comparand1 itself, respectively, if it has no language tag) is respectively less than, equal to, or greater than the value of the string-part of $comparand2 (or $comparand2 itself, respectively, if it has no language tag). The collation used by the invocation of this function is determined according to the rules in Section 7.3.1 of the XPath and XQuery functions specification [XPathFunc].

The first version of this function backs up the XQuery operators "eq", "ne", "gt", "lt", "le", and "ge" on rdf:PlainLiteral values.

The two functions may be viewed as declared XQuery functions with the following definitions:

declare function  rtfn:compare( $comparand1 as rdf:text?, $comparand2 as rdf:text? ) as xs:integer?
 {
  return
    if ( fn:compare (  rtfn:lang-from-text( $comparand1 ),   rtfn:lang-from-text( $comparand2 ) ) = 0 ) then
       fn:compare (  rtfn:string-from-text( $comparand1 ) ,  rtfn:string-from-text( $comparand2 ) )
 }

declare function  rtfn:compare( $comparand1  as rdf:text?, $comparand2 as rdf:text? $collation as xs:string ) as xs:integer?
 {
  return
    if ( fn:compare ( fn:lang-from-text( $comparand1 ),   rtfn:lang-from-text( $comparand2 ) ) = 0 ) then
       fn:compare (  rtfn:string-from-text( $comparand1 ) ,  rtfn:string-from-text( $comparand2 ), $collation)
 }

5.3 Other Functions on rdf:PlainLiteral Data Values

5.3.1 rtfn:length

 rtfn:length($arg as rdf:PlainLiteral) as xs:integer

Summary: returns the number of characters in the string part s if $arg is an rdf:text data value of the form ( s, l ) or a string value s, respectively. If $arg is not of type rdf:text, this function raises type error err:FORG0006.

This function may be viewed as a declared XQuery function with the following definition:

declare function  rtfn:text-length($arg as rdf:text?) as xs:integer
 {
  return
     fn:string-length (  rtfn:string-from-text( $arg ) )
 }

5.3.2 rtfn:matches-language-range

 rtfn:matches-language-range($arg as rdf:PlainLiteral?, $range as xs:string) as xs:boolean

Summary: This function is only defined if $arg is a sequence of length 0 or 1 of literals of type rdf:text and $range is of type xs:string; if the parameters do not satisfy these typing conditions, the function raises a type error err:FORG0006. If the typing conditions are fulfilled, the function returns true in case $arg is an rdf:text data value of the form ( s, l ) with l a language tag that matches the extended language range $range as specified by the extended filtering algorithm for "Matching of Language Tags" [BCP-47]; otherwise, it returns false. This means that the function returns false if the argument is a string rdf:text data value. An empty input sequence is treated as a rdf:text data value consisting of the empty string, and accordingly on such input this function also returns false.

6 Acknowledgments

The RIF WG and the OWL WG made parallel efforts to support strings written in different languages. This specification is the outcome of a collaboration between the two groups, and it is based on the work on the rif:text datatype on the RIF side and the owl:internationalizedString datatype on the OWL side. A short description of the design process is available here.

7 References

[RFC 2119]

RFC 2119: Key words for use in RFCs to Indicate Requirement Levels. Network Working Group, S. Bradner. Internet Best Current Practice, March 1997.

[RFC 3986]

RFC 3986 - Uniform Resource Identifier (URI): Generic Syntax. T. Berners-Lee, R. Fielding, and L. Masinter, IETF, January 2005.

[RFC 4647]

RFC 4647 - Matching of Language Tags. A. Phillips and M. Davis, IETF, September 2006.

[UNICODE]

The Unicode Standard. Unicode The Unicode Consortium, Version 5.1.0, ISBN 0-321-48091-0, as updated from time to time by the publication of new versions. (See http://www.unicode.org/unicode/standard/versions for the latest version and additional information on versions of the standard and of the Unicode Character Database)."

[ISO/IEC 10646]

ISO/IEC 10646-1:2000. Information technology  Universal Multiple-Octet Coded Character Set (UCS)  Part 1: Architecture and Basic Multilingual Plane and ISO/IEC 10646-2:2001. Information technology  Universal Multiple-Octet Coded Character Set (UCS)  Part 2: Supplementary Planes, as, from time to time, amended, replaced by a new edition or expanded by the addition of new parts. [Geneva]: International Organization for Standardization. ISO (International Organization for Standardization).

[BCP 47]

BCP-47 - Tags for Identifying Languages. A. Phillips, M. Davis, eds., IETF, September 2006, http://www.rfc-editor.org/rfc/bcp/bcp47.txt.

[RDF]

Resource Description Framework (RDF): Concepts and Abstract Syntax. Graham Klyne, Jeremy J. Carroll, and Brian McBride, eds., W3C Recommendation 10 February 2004.

[RDF Semantics]

RDF Semantics. Patrick Hayes, ed., W3C Recommendation 2004

[XML]

Extensible Markup Language (XML) 1.0 (Fifth Edition). Tim Bray, Jean Paoli, C. M. Sperberg-McQueen, Eve Maler, and Fran?ois Yergeau, eds., W3C Recommendation 26 November 2008.

[XML Namespaces]

Namespaces in XML 1.0 (Second Edition). Tim Bray, Dave Hollander, Andrew Layman, Richard Tobin, eds., W3C Recommendation 16 August 2006.

[XML Schema Datatypes]

W3C XML Schema Definition Language (XSD) 1.1 Part 2: Datatypes. D. Peterson, S. Gao, A. Malhotra, C. M. Sperberg-McQueen, H. S. Thompson, eds., W3C Working Draft 30 January 2009.

[XPath20]

XML Path Language (XPath) 2.0. Anders Berglund, Scott Boag, Don Chamberlin, Mary F. Fern?ndez, Michael Kay, Jonathan Robie, and J?r?me Sim?on, eds. W3C Recommendation 23 January 2007.

[XPathFunc]

XQuery 1.0 and XPath 2.0 Functions and Operators. Ashok Malhotra, Jim Melton, and Norman Walsh, eds. W3C Recommendation 23 January 2007.

[BIDI]

Unicode controls vs. markup for bidi support. Richard Ishida, W3C Consortium, November 22 2007.

[RUBY]

Ruby Annotation. Marcin Sawicki, Michel Suignard, Masayasu Ishikawa, Martin D?rst, and Tex Texin, eds. W3C Recommendation 31 May 2001.